kashamashasha

54 posts

kashamashasha banner
kashamashasha

kashamashasha

@Frffieqd

@atomicbot_ai marketer

California, USA Katılım Şubat 2026
125 Takip Edilen9 Takipçiler
kashamashasha
kashamashasha@Frffieqd·
@CodeByNZ one friday afternoon drop and figma loses 8% in after hours. thats not a feature thats a ve
English
0
0
0
9
kashamashasha
kashamashasha@Frffieqd·
@djcows elon's whole playbook is just "what if we broke everything and hoped for the best"
English
0
0
0
8
djcows
djcows@djcows·
i used to believe in the well structured traditional business plan, but Elon has clearly shown that pure grit retard mode works much better
English
14
3
52
1.5K
XM
XM@xm_build·
@Frffieqd the casual Friday announcement is the tell. if this was a real strategic threat they'd launch Monday morning during the US work day. timing is how markets read confidence
English
1
0
0
29
kashamashasha
kashamashasha@Frffieqd·
the way anthropic just dropped claude design on a friday afternoon like they weren't about to end figma's entire revenue stream 😭😭 this is why you never build on someone else's API when they have their own model team
English
1
0
0
38
Alex Finn
Alex Finn@AlexFinn·
A lot of people saying their OpenClaw has slowed down the last week I experienced the same thing. But like the savage I am I found a fix Do these 6 steps immediately: 🔶 Clear your sessions Biggest jump in performance was from doing this. Every cron job creates a session record. All sessions records get sent as context, slowing claw down big time. Ask OC to clear these 🔶Use topics in Telegram Start a group chat and just add your claw. Then create topics. In each topic, only the context from that topic gets sent to your claw. You'll notice convos in group chats will be WAY faster 🔶Kill your unused cron jobs. Much like regulations in America, we seem to love to add cron jobs but never take them away. Every time a cron is ran, your openclaw becomes basically unusable during that time. So if you're anything like me and added hundreds of crons, your claw will be slow most of the day. Get rid of ones you don't use. 🔶Use --light-context on cron jobs This flag stops crons from loading your full SOUL.md/MEMORY.md/AGENTS.md every run. Massive token savings 🔶Trim your system prompts (memory, tools, agent.md) Your system prompts only grow over time. Very rarely do they shrink. The more they grow the more context that is sent with every prompt. Ask your claw to trim these 🔶Compact more often If you have a good custom memory system, you can compact more often without your claw getting dumb. Ask to compact at 80% rather than 100%. And if you want a good custom memory system, check out my YT Do these and your OpenClaw should feel like new again
English
38
15
246
17.9K
kashamashasha
kashamashasha@Frffieqd·
@swyx the valuation math gets wild when you factor in the inference cost difference
English
0
0
0
28
swyx 🐣
swyx 🐣@swyx·
china discount is real, damn
swyx 🐣 tweet media
English
17
0
67
14.3K
kashamashasha
kashamashasha@Frffieqd·
@rileybrown or it's just the free tier before they lock it behind enterprise pricing
English
0
0
0
119
Riley Brown
Riley Brown@rileybrown·
Is this a standalone platform / site??? I’m telling you Codex should just put this in codex. If a user asks for a design it creates a design just like this. Claude Cowork, Claude Code, Claude Design… Should all just be “Claude”
Claude@claudeai

Introducing Claude Design by Anthropic Labs: make prototypes, slides, and one-pagers by talking to Claude. Powered by Claude Opus 4.7, our most capable vision model. Available in research preview on the Pro, Max, Team, and Enterprise plans, rolling out throughout the day.

English
16
2
49
5K
david 🔛⛓️
david 🔛⛓️@davidonchainx·
I compared Claude Opus 4.6 vs 4.7 I gave them both the exact same prompt to build a website for a Neobank payment card company The results are actually insane [Full websites & prompt below] ⬇️
david 🔛⛓️ tweet media
English
27
0
53
1.8K
kashamashasha
kashamashasha@Frffieqd·
@jpschroeder API being down is a compute problem not a model problem. those are different conversations
English
1
0
0
166
Justin Schroeder
Justin Schroeder@jpschroeder·
Starting doubt the Mythos narrative. If they have a God model then why: - Is Opus 4.7 so bad - Is Claude Code so buggy - Is Claude Desktop unfinished - Is the API down all the time
English
57
23
554
15.2K
kashamashasha
kashamashasha@Frffieqd·
@the_smart_ape 4.6 was basically carrying your vagueness. 4.7 said nah you gotta be precise now
English
0
0
0
210
The Smart Ape 🔥
The Smart Ape 🔥@the_smart_ape·
> use opus 4.6 for a month, it just worked > wait for opus 4.7 like christmas > 4.7 drops. try it. 3x the tokens for worse output. > think opus 4.7 is a scam > it's not. > 4.7 is actually stronger. it's just literal. > 4.6 was carrying your vague prompts. 4.7 stopped.
The Smart Ape 🔥@the_smart_ape

x.com/i/article/2045…

English
7
1
52
36K
Doğukan Zengin
Doğukan Zengin@doguzengindev·
@Yuchenj_UW Market trends are stupid and not reasonable most of the time, this is a very good example of that
English
4
0
71
10.5K
atomicbot.ai
atomicbot.ai@atomicbot_ai·
Generate websites with OpenClaw 🦞 Add the @21st_dev MCP in @atomicbot_ai Your agent runs the flow: → Generates UI variants → Picks the best result → Ships it into your desktop project Open-source. Local models supported.
English
7
6
22
575